чoe?ddlZddlZddlmZddlZddlZddlZddlmZddlm Z ddl m Z m Z m Z ddlZddlmZddlmZddlZ ejeZejeZn#e$rdZdZYnwxYwdd lmZmZmZmZd Zd Z gd Z!d Z"dZ#d Z$dZ%dZ&dZ'd)dZ(d*dZ)dZ*dZ+dZ,dZ-dZ.dZ/dZ0dZ1dZ2dZ3d Z4d+d!Z5d,d#Z6d$Z7ed-d&Z8d'Z9d(Z:dS).N)contextmanager) defaultdict)getenv)abspathdirnameexists)md5)open)vlen)meanmedianminmaxDaniel McDonaldz5Copyright 2011-2020, The BIOM Format Development Team)rzJai Ram Rideoutz Greg Caporasoz Jose ClementezJustin KuczynskiuJorge Cañardo AlastueyBSDzhttp://biom-format.orgzdaniel.mcdonald@colorado.edu)z2.1.15sampleFc#<K ||||V)aIndefinitely generate random subsamples Parameters ---------- table : Table The table to subsample from n : int The size of the subsample axis : {'sample', 'observation'}, optional The axis to operate on, defaults to 'sample'. by_id : bool, optional If `True`, operate on IDs, if `False`, operate on values. Defaults to `False`. Returns ------- GeneratorType Each subsequent call to .next() will yield a new randomly subsampled table Examples -------- Randomly subsample the samples in the example_table 10 times: >>> from biom import example_table >>> gen = generate_subsamples(example_table, 2, by_id=True) >>> observed_ids = [] >>> for _, table in zip(range(100), gen): ... observed_ids.append(tuple(table.ids())) >>> print(sorted(set(observed_ids))) [('S1', 'S2'), ('S1', 'S3'), ('S2', 'S3')] ) subsample)tablenaxisby_ids ^/mounts/lovelace/software/anaconda3/envs/kraken-biom/lib/python3.11/site-packages/biom/util.pygenerate_subsamplesr/s-F.ooau-----.c4|dSd|dd|ddS)zReturns the current Biom file format version. Parameters ---------- version : tuple a tuple containing the version number of the biom table Nz#Biological Observation Matrix 1.0.0zBiological Observation Matrix r.rz.0)versions rget_biom_format_version_stringr"Vs044K KKWQZKKKKrctS)z5Returns the current Biom file format description URL.)__url__r rrget_biom_format_url_stringr%ds Nrc v|r6ttttt|SgS)aAPerforms the reverse of zip, i.e. produces separate lists from tuples. items should be list of k-element tuples. Will raise exception if any tuples contain more items than the first one. Conceptually equivalent to transposing the matrix of tuples. Returns list of lists in which the ith list contains the ith element of each tuple. Note: zip expects *items rather than items, such that unzip(zip(*items)) returns something that compares equal to items. Always returns lists: does not check original data type, but will accept any sequence. This function is ported from PyCogent (http://www.pycogent.org). PyCogent is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). )listmapzip)itemss runzipr+is4, Cc4;;/00111 rcxg}|D]4} ||#||Y2xYw|S)ahRemoves one level of nesting from items. items can be any sequence, but flatten always returns a list. This function is ported from PyCogent (http://www.pycogent.org). PyCogent is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). )extendappend)r*resultis rflattenr1sXF   MM!      MM!      Ms7ct|} tjd|}n+#t$rtjd|d}YnwxYwt t |D]W}||r@||ddvr0d||vrt }nt}d|||f||<Jd||f||<X||fS)aProvides normalized version of item for sorting with digits. This function is ported from QIIME (http://www.qiime.org) and is based on: http://lists.canonical.org/pipermail/kragen-hacks/2005-October/000419.html QIIME is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). z(\d+(?:\.\d+)?)r 0123456789rr)strresplit TypeErrorrangelenfloatint)itemchunksiinumtypes r _natsort_keyr@s t99D7,d33 777,d1g667CKK  ) ) ": )&*Q-<77fRj  WWVBZ001F2JJVBZF2JJ D>s'%AAcZt|}|t|S)aSort a sequence of text strings in a reasonable order. This function is ported from QIIME (http://www.qiime.org) and is based on: http://lists.canonical.org/pipermail/kragen-hacks/2005-October/000419.html QIIME is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). )key)r'sortr@)seqalists rnatsortrFs( IIE JJzindex_list..s 5 5 5cC 5 5 5r) enumerate)r<s r index_listrRs 5 5Yt__ 5 5 55rcg}t}||dztd}|r||td}|r|dz}||g}|D]3}t|r"|t |4t |S)aLReturns biom-format configuration read in from file. This function is ported from QIIME (http://www.qiime.org), previously named load_qiime_config. QIIME is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). z/support_files/biom_configBIOM_CONFIG_FPHOMEz /.biom_config)get_biom_project_dirr.rrr parse_biom_config_files)biom_config_fpsbiom_project_dirbiom_config_env_fphome_dirbiom_config_home_fpbiom_config_filesbiom_config_fps rload_biom_configr_sO+--+.JJKKK 01131222f~~H4&82333);; . ! ! ;  $ $T.%9%9 : : : "#4 5 55rctt}t|}t|}t|S)aHReturns the top-level biom-format directory. This function is ported from QIIME (http://www.qiime.org), previously named get_qiime_project_dir. QIIME is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). )r__file__r) current_fp current_dirs rrVrVs?""J*%%K+&&K ;  rcd}t|}|D]5} |t|&#t$rY2wxYw|S)aParses files in (ordered!) list of biom_config_files. The order of files must be least important to most important. Values defined in earlier files will be overwritten if the same values are defined in later files. This function is ported from QIIME (http://www.qiime.org), previously named parse_qiime_config_files. QIIME is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). cdSNr r rr return_nonez,parse_biom_config_files..return_none str)rupdateparse_biom_config_fileOSError)r]rgresultsbiom_config_files rrWrWs{+&&G-  NN12BCC D D D D    D  Ns"; AAci}|D]n}|}|r|dr.|}|d}d|ddpd}|||<o|S)aAParses lines in a biom_config file. This function is ported from QIIME (http://www.qiime.org), previously named parse_qiime_config_file. QIIME is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). #r rN)strip startswithr6join)rlr/linefieldsparam_id param_values rririsF ' 'zz|| ts++  !9hhvabbz**2d &x Mrci}|D]H\}}}|r|dk||<$t|||<It|}t |dkrdddd|fSt |t|t|t||fS)a@Return summary statistics on per-sample observation counts table: a BIOM table object binary_counts: count the number of unique observations per sample, rather than the sum of the total counts (i.e., counts are qualitative rather than quantitative) This function is ported from QIIME (http://www.qiime.org), previously named compute_seqs_per_library_stats. QIIME is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). r) itersumr:r'valuesr9rrr r )r binary_counts sample_counts count_vector sample_idmetadatacountss rcompute_counts_per_sample_statsr0sM-2ZZ\\AA) i  A(4(9'>'>'@'@M) $ $',\-=-=-?-?'@'@M) $ $ -&&(( ) )F 6{{a1aM**F F vV   rcnd}t}tdr j}|}n>ttr fd}|}d}nt dt z|r7||}|r(||d|7| S)aComputes an md5 sum without loading the file into memory This method is based on the answers given in: http://stackoverflow.com/questions/1131220/get-md5-hash-of-a-files-without-open-it-in-python This function is ported from PyCogent (http://www.pycogent.org). PyCogent is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). TreadcR |S#t$rYdSwxYwrf)pop IndexError)r0 open_files rfzsafe_md5..fcs<  }}Q'''   tt s  &&rzHsafe_md5 can only handle a file handle or list of lines but recieved %r.utf-8) r hasattrr isinstancer'r7typerhencode hexdigest)r block_sizedatar/ data_getter data_getter_irs` rsafe_md5rOs D UUFy&!! Dn " It $ $ D        137 ??CDD D 0{=))  0 MM$++g.. / / / 0     rct|d5}|ddkcdddS#1swxYwYdS)aChecks the first two bytes of the file for the gzip magic number If the first two bytes of the file are 1f 8b (the "magic number" of a gzip file), return True; otherwise, return false. This function is ported from QIIME (http://www.qiime.org). QIIME is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). rbrsNr rfprs ris_gziprus b$(1vvayyK'(((((((((((((((((( 7;;rc#zK|dvrtd|ztjtjd}|}|dvr5t j|dkrtd|z|dvr+tj |rtj }|d krd n|}n|d kr tj }|dvrt|r d }|d vrdn|}n)|dvr%t|drd}|||} |V|dS#|wxYw)atWrapper to allow opening of gzipped or non-compressed files Read or write the contents of a file Parameters ---------- file_fp : file path or pathlib.Path permission : str, {'r', 'w', 'wb', 'rb', 'U'} Returns ------- [h5py.File, file, gzip.GzipFile] Notes ----- If the file is binary, be sure to pass in a binary mode (append 'b' to the mode); opening a binary file in text mode (e.g., in default mode 'U') will have unpredictable results. If h5py is available on the system, you cannot use biom_open to create a writable ASCII file handle. You can use it to create writable GZIP handles and HDF5 handles, however. This function is ported from QIIME (http://www.qiime.org), previously named qiime_open. QIIME is a GPL project, but we obtained permission from the authors of this function to port it to the BIOM Format project (and keep it under BIOM's BSD license). Raises ------ RuntimeError If the user tries to parse an HDF5 file without having h5py installed. ValueError If the user tries to read an empty file. )rwUrwbzUnknown mode: %sr)encoding>rrrrz*The file '%s' is empty and can't be parsed)rrrrrrcXtjdt||SNr)codecs getreader gzip_openrmodes ropenerzbiom_open..openers',6#G,,Yr4-@-@AA Ar)rrr)rrz.gzc\tjdt||dSr)r getwriterrrs rrzbiom_open..openers- %F W % %iD&9&9 : : : : :rN)rj functoolspartialior ospathgetsize ValueErrorh5pyis_hdf5Filerr4endswithclose)r permissionrrrs r biom_openrs|L444(:5666  rw 9 9 9F D  7??2  ! # #IBNOO O DL$4$4 C''ssZ  GBKK B B B!Z//ttZ   R!1!1%!8!8  ; ; ; r4A   s D$$D:c"tjtjdd}tjtj|}tj|d|}|S)aTReturn path to filename ``fn`` in the data folder. During testing it is often necessary to load data files. This function returns the full path to files in the ``data`` subfolder. Parameters ---------- fn : str File name. Returns ------- str Inferred absolute path to the test data for the module where ``get_data_path(fn)`` is called. Notes ----- The requested path may not point to an existing file, as its existence is not checked. This method was adapted from scikit-bio, specifically `skbio.util.testing`. r test_data)inspectgetouterframes currentframerrrrrr)fncallers_filenamer data_paths r get_data_pathrsg6-g.B.D.DEEaHK 7??27??+;<< = =D T;33I rct|d5}|ddkcdddS#1swxYwYdS)zGuess if file is HDF5. Parameters ---------- fn : str File name Returns ------- bool Whether the file is an HDF5 file rsHDF  Nrrs r is_hdf5_filers b$11vvayy00111111111111111111r)rFrf)F)r)r);rr contextlibrrrr collectionsrros.pathrrrr5hashlibr gzipr rr special_dtyper4 H5PY_VLEN_STRH5PY_VLEN_UNICODE ImportErrornumpyr r rr __author__ __copyright__ __credits__ __license__r$__maintainer__ __email____format_version__ __version__rr"r%r+r1r@rFrJrRr_rVrWrirrrrrrr rrrs %%%%%% ######,,,,,,,,,, """""" &D&C000M**444M)(((((((((((  G OOO  "" *  $.$.$.$.N L L L L 8&:   %%% 666 666:   *6,>####L ( ( (CCCCLB11111s"A)) A54A5